Dear friends, in a long ago have helped me to correct the dc offset my rotel ra-970BX amplifier. has remained at 20mV positive channels and 20 mv negative on the other channel. Satisfied as it was +70 and - 60 mv.
Now I want your feedback to improve its sound (which is very good).
My idea is to replace the electrolytic capacitors power supply (already old) Originales dicen ROTEL. Me he comunicado con empresa y no envian . Cual reemplazo?.And replace electrolytic input to the preamplifier IC; the output preamplifier and input power (10 uF, 10 uF and 4.7 uF) by Jantzen Z Caps Silver MKP.
Question, can this improve purity of sound ?. I have with this amp, Marantz cd5400, capsule Audiotechnica magnetic phono preamp AT440MLa with Rod Elliot circuit, cables Van den Hull and B & W DM602 speakers.
* Greetings and already thanks. Project SolidPhono almost assembly TNT. (sorry for my English)
Now I want your feedback to improve its sound (which is very good).
My idea is to replace the electrolytic capacitors power supply (already old) Originales dicen ROTEL. Me he comunicado con empresa y no envian . Cual reemplazo?.And replace electrolytic input to the preamplifier IC; the output preamplifier and input power (10 uF, 10 uF and 4.7 uF) by Jantzen Z Caps Silver MKP.
Question, can this improve purity of sound ?. I have with this amp, Marantz cd5400, capsule Audiotechnica magnetic phono preamp AT440MLa with Rod Elliot circuit, cables Van den Hull and B & W DM602 speakers.
* Greetings and already thanks. Project SolidPhono almost assembly TNT. (sorry for my English)
Don't bother with these huge "audiophile" film caps, which have parasitics to match their size. With some bad luck, they make some very nice antennas, leading to oscillation. Better leave these for speaker crossovers and stick with standard Panasonic and WIMA film caps if need be.
If you do feel like replacing coupling caps (and feedback caps, like C688), look for some nice "low-leakage" type electrolytics (Elna RLB, Vishay / BCComponents 013 RLC, Nichicon KL).
Some Rotels could have issues with carbon film resistors, but it looks like parts quality in this one is quite good, with a lot of metal films. I would thus also expect most electrolytics to be quite healthy still, though ones in particularly warm spots may appreciate being replaced. (Do you hear any hum?)
The one major sonic bottleneck in this amp would have to be its rather high noise level. This should be mainly a result of using a relatively noisy opamp after the volume pot, the AD827, actually a video opamp and rated at 15 nV/sqrt(Hz) in terms of voltage noise. (Rotel used to do this a lot.) Try an LM4562 family part, preferably the LME49860 with its higher voltage rating (+/- 16.8 V nominal is pushing the spec of +/-17 V max for LM4562/LME49720, though it should still work). Unlike dubious parts changes, this one should be readily audible!
If you do feel like replacing coupling caps (and feedback caps, like C688), look for some nice "low-leakage" type electrolytics (Elna RLB, Vishay / BCComponents 013 RLC, Nichicon KL).
Some Rotels could have issues with carbon film resistors, but it looks like parts quality in this one is quite good, with a lot of metal films. I would thus also expect most electrolytics to be quite healthy still, though ones in particularly warm spots may appreciate being replaced. (Do you hear any hum?)
The one major sonic bottleneck in this amp would have to be its rather high noise level. This should be mainly a result of using a relatively noisy opamp after the volume pot, the AD827, actually a video opamp and rated at 15 nV/sqrt(Hz) in terms of voltage noise. (Rotel used to do this a lot.) Try an LM4562 family part, preferably the LME49860 with its higher voltage rating (+/- 16.8 V nominal is pushing the spec of +/-17 V max for LM4562/LME49720, though it should still work). Unlike dubious parts changes, this one should be readily audible!
Last edited:
Many thanks sgrossklass after reading your advice, you convinced me to leave my Rotel factory as it is. Actually I have taken for advice from others who say there is improvement in detail using such capacitor. Thank you very much.
Maybe try changing the operational amp for your RECOMMEND me. For now I leave it as it is, it sounds great.
Maybe try changing the operational amp for your RECOMMEND me. For now I leave it as it is, it sounds great.
Sgrossklass. Sorry for the question. The operational amp AD827 has 300V / usec slew rate and the LME49720 has 20 V / ps. This influences the resulting dynamic ?, or is "flat".
If so some noise prefer to stay with little dynamic. Greetings.
If so some noise prefer to stay with little dynamic. Greetings.
The maximum slew rate to occur in a sinusoidal signal of peak amplitude A and frequency f is
max(SR) = max(dV/dt) = 2 pi A f
For A = (16.8 V - 1.2 V) = 15.6 V and f = 20 kHz, we are getting
max(SR) = 1.96 V/µs.
So a part with, say, 6 V/µs ought to be just fine. For a signal that's as large as supply voltages will allow here, at 20 kHz. Now
1. the opamp will never need to swing anything even close to 15 Vp = 11 Vrms in this circuit. The power amp will do 80 W into 8 Ohms with a following wind (that's 25.3 Vrms), while having a voltage gain of 22.5. Maximum input level required hence is about 1.1 Vrms. So realistically, 0.6 V/µs would do fine. Which is µA741 level.
2. How often do you get a 20 kHz full-scale signal in practice? Maybe when running a CCIF intermodulation test, but in music this sort of this is exceptionally rare.
I once helped a guy tweak his RA-980BX, which is similar. He bravely put in an LME49720 (I really would prefer the '49860 there - rails are +/-17.3 V nominal in this one), and reported success. Thankfully rail bypassing in these units is quite good. The good thing about the '970BX is that you don't even need to modify the feedback network, just swap out the opamp itself.
max(SR) = max(dV/dt) = 2 pi A f
For A = (16.8 V - 1.2 V) = 15.6 V and f = 20 kHz, we are getting
max(SR) = 1.96 V/µs.
So a part with, say, 6 V/µs ought to be just fine. For a signal that's as large as supply voltages will allow here, at 20 kHz. Now
1. the opamp will never need to swing anything even close to 15 Vp = 11 Vrms in this circuit. The power amp will do 80 W into 8 Ohms with a following wind (that's 25.3 Vrms), while having a voltage gain of 22.5. Maximum input level required hence is about 1.1 Vrms. So realistically, 0.6 V/µs would do fine. Which is µA741 level.
2. How often do you get a 20 kHz full-scale signal in practice? Maybe when running a CCIF intermodulation test, but in music this sort of this is exceptionally rare.
I once helped a guy tweak his RA-980BX, which is similar. He bravely put in an LME49720 (I really would prefer the '49860 there - rails are +/-17.3 V nominal in this one), and reported success. Thankfully rail bypassing in these units is quite good. The good thing about the '970BX is that you don't even need to modify the feedback network, just swap out the opamp itself.
Vicente (Dyaudio.es) is a very good spanish technician.
Rotel RA-970 MKII (Mod by Diyaudio.es)
-> Diyaudio.es: Rotel RA-970 MKII (Mod by Diyaudio.es)
-> USB Isolator: iFi iUSB vs TeraDak U9VA + Teralink ADuM4160. Filtros Schaffner. Lampizator: AC filter DIY ESA SILK. Estabilizador / regulador de tensión / voltaje. Isolation / Balanced transformer. DC Blocker / Blocking. PC SILENCIOSO en Aussar. - P
I like very very very much EPCOS B41505A7189M000 CAP, ALUM ELECT, 18000UF, 35V, CAN into my second DC blocking:
-> USB Isolator: iFi iUSB vs TeraDak U9VA + Teralink ADuM4160. Filtros Schaffner. Lampizator: AC filter DIY ESA SILK. Estabilizador / regulador de tensión / voltaje. Isolation / Balanced transformer. DC Blocker / Blocking. PC SILENCIOSO en Aussar. - P
Rotel RA-970 MKII (Mod by Diyaudio.es)
-> Diyaudio.es: Rotel RA-970 MKII (Mod by Diyaudio.es)
Or EPCOS (TDK) capacitors, like me:
Trabajo realizado: Cambio Condensadores 10.000uf/63V Nichicon KG (Gold Tune). Cambio operacionales tono y volumen 2 x LME49860
-> USB Isolator: iFi iUSB vs TeraDak U9VA + Teralink ADuM4160. Filtros Schaffner. Lampizator: AC filter DIY ESA SILK. Estabilizador / regulador de tensión / voltaje. Isolation / Balanced transformer. DC Blocker / Blocking. PC SILENCIOSO en Aussar. - P
I like very very very much EPCOS B41505A7189M000 CAP, ALUM ELECT, 18000UF, 35V, CAN into my second DC blocking:
-> USB Isolator: iFi iUSB vs TeraDak U9VA + Teralink ADuM4160. Filtros Schaffner. Lampizator: AC filter DIY ESA SILK. Estabilizador / regulador de tensión / voltaje. Isolation / Balanced transformer. DC Blocker / Blocking. PC SILENCIOSO en Aussar. - P
Who would have thought, the LME49860 again. 🙂
Actually performance of the circuit should already improve markedly when swapping out the AD827 for the NJM2114 in the tone control and vice versa.
In stock form, a low-noise hottie like the NJM2114 is pretty much a total waste there because it's swamped by noise from the preceding AD827 voltage amplifier with a gain of 16.5 dB. Swap them around, and things start making a lot more sense.
I am under the impression that the Rotel guys liked to use the trendy opamp du jour without having much of a clue as to why. Not like audio designers wouldn't do this today...
Actually performance of the circuit should already improve markedly when swapping out the AD827 for the NJM2114 in the tone control and vice versa.
In stock form, a low-noise hottie like the NJM2114 is pretty much a total waste there because it's swamped by noise from the preceding AD827 voltage amplifier with a gain of 16.5 dB. Swap them around, and things start making a lot more sense.
I am under the impression that the Rotel guys liked to use the trendy opamp du jour without having much of a clue as to why. Not like audio designers wouldn't do this today...
Last edited:
PS: Trying to match the NJM2114 input noise voltage vs. source impedance graph in the datasheet, I'm getting about en = 3.5 nV/sqrt(Hz) and in = 1.0 pA/sqrt(Hz)... The en(f) graph looks more like 4.5 nV/sqrt(Hz), but that included an extra ~200 ohms, which should give about 4.0 total.
Anyway, it looks like the part should have a bit more voltage noise but less current noise than the LM4562/LME49720/49860. Simulated results when used for IC501 in the existing RA-970BX circuit are 1.41 µV of output noise for the NJM2114 vs. 1.53 µV for the NatSemi part, at zero volume. Lower feedback network resistor values could reverse positions, but the NatSemi parts would remain at a disadvantage of up to 3 dB when turning up the volume.
Hence I would recommend using the existing NJM2114 (IC502) in place of IC501, with an LME49860 to go into the now-vacant IC502 position.
Anyway, it looks like the part should have a bit more voltage noise but less current noise than the LM4562/LME49720/49860. Simulated results when used for IC501 in the existing RA-970BX circuit are 1.41 µV of output noise for the NJM2114 vs. 1.53 µV for the NatSemi part, at zero volume. Lower feedback network resistor values could reverse positions, but the NatSemi parts would remain at a disadvantage of up to 3 dB when turning up the volume.
Hence I would recommend using the existing NJM2114 (IC502) in place of IC501, with an LME49860 to go into the now-vacant IC502 position.
Sgrossklass,friend beaause of my difficulty whith English i misunderstood operational amplifier wich would be best for my Rotel. I am willing to make the change but i did not realize at that. Thanks for the explanation of slew rate. Exelent.
I hope I'm getting through the language barrier halfway well. My Spanish is near-inexistent, I'm afraid (I do understand some but wouldn't be able to write much).
I just noticed I must have made a mistake in the voltage noise calculation earlier, probably my noise calculator fooled me again because it expects input in kOhms, and I must have forgotten to enter the 681 ohms as "0.681". 1.xx µV of noise output would have been too good to be true.
Correct values for IC501 noise output would be (20 kHz bandwidth):
NJM2114: 4.73 µV @ min volume, 5.45 µV @ -40 dB position, 14.37 µV max
LME49860: 4.32 µV @ min volume, 5.10 µV @ -40 dB, 14.24 µV max
AD827: 14.68 µV @ min volume, 14.93 µV @ -40 dB, 19.99 µV max
As you can see, the difference between NJM2114 and LME49860 is quite small (about 1 dB or less), but both result in a considerable improvement over the stock AD827 - more than 9 dB less noise. The owner of the '980BX stated that hiss went from plainly audible at normal listening distance to only detectable with your ear close to the tweeters.
I just noticed I must have made a mistake in the voltage noise calculation earlier, probably my noise calculator fooled me again because it expects input in kOhms, and I must have forgotten to enter the 681 ohms as "0.681". 1.xx µV of noise output would have been too good to be true.
Correct values for IC501 noise output would be (20 kHz bandwidth):
NJM2114: 4.73 µV @ min volume, 5.45 µV @ -40 dB position, 14.37 µV max
LME49860: 4.32 µV @ min volume, 5.10 µV @ -40 dB, 14.24 µV max
AD827: 14.68 µV @ min volume, 14.93 µV @ -40 dB, 19.99 µV max
As you can see, the difference between NJM2114 and LME49860 is quite small (about 1 dB or less), but both result in a considerable improvement over the stock AD827 - more than 9 dB less noise. The owner of the '980BX stated that hiss went from plainly audible at normal listening distance to only detectable with your ear close to the tweeters.
Sgosscklass, I will buy on Ebay 10000/63 KG Nichicon capacitors; LME49860 and operational. I need just 1 since I use the amplifier in flat, bypassing the tone controls.
You could replace the diode bridge power supply other high speed and higher amperage ?. Greetings and thanks.
You could replace the diode bridge power supply other high speed and higher amperage ?. Greetings and thanks.
Friends, before buying on Ebay, I would confirm that I am right.
-Nichicon KG Gold Tune I 10000/63v electrolitic capacitor x 2
-LME49860 Operational amp
Thanks and sorry for the insistence.
-Nichicon KG Gold Tune I 10000/63v electrolitic capacitor x 2
-LME49860 Operational amp
Thanks and sorry for the insistence.
For the electrolytics:
Verify what sort of diameter and number of pins you need and which maximum height would be possible. Most nasty surprises with electrolytics are of mechanical nature. Other companies have nice caps too, like the Panasonic slit foil ones.
Don't buy from shady fly-by-night Chinese dealers. Have a close look at the seller's track record. While I don't think there are any LME49860 fakes around (yet), relabelling chips seems to be easy enough.
Verify what sort of diameter and number of pins you need and which maximum height would be possible. Most nasty surprises with electrolytics are of mechanical nature. Other companies have nice caps too, like the Panasonic slit foil ones.
Don't buy from shady fly-by-night Chinese dealers. Have a close look at the seller's track record. While I don't think there are any LME49860 fakes around (yet), relabelling chips seems to be easy enough.
Sgrossklass Thanks, I will consider the physical measurements of electrolytic. As for the Chinese vendors, almost have no choice. Everything sold is 99% Chinese internet. I hope they are not fakes because each electrolytic costs U $ S 22 °°.
Regarding the LME could not say the same. It costs U $ S 10 °° par.
Greetings and thank you very much again. After the upgrade I tell them.
Regarding the LME could not say the same. It costs U $ S 10 °° par.
Greetings and thank you very much again. After the upgrade I tell them.
You can bet on the electrolytics on Ebay being fake. Their high prices are just there to throw you off.
Welcome Friend. Unfortunately I just made the purchase. U $ S 26 °°. But if you suggest some place to do shopping with cofianza, I will keep these purchased on Ebay and I'll make another purchase. Thanks and hope your suggestion
Friends, I have received from China, bought by Ebay, Nichicon electrolytic capacitors. I measured the capacity loading them from a power source with a 100k resistor 1% (98,000 ohms).
Then I calculated the time constant formula and then I confirmed with Digi-key calculator.
The calculations give me 12495 and 12480 uF.
I think they are not fakes, but sending photos for you to THINK.
Then I calculated the time constant formula and then I confirmed with Digi-key calculator.
The calculations give me 12495 and 12480 uF.
I think they are not fakes, but sending photos for you to THINK.
Dude, I'll gladly give you the same help that at some point I have dadi in this forum.
The Rotel RA 970BX is an excellent amplifier and should work with the best possible parameters.
I sent the post, dating back three years, but only this year I was encouraged to make the change that had advised me. Reduce the R603 / 604 resistors advised by Wahab.
http://www.diyaudio.com/forums/solid-state/213432-dc-offset-rotel.html
Finally, after three years I have decided to make a change that I had suggested.
Is to bring R 604 and R603 47k to 23k5 (placing another resistor in parallel 47k 1% metal film).
Result: -0.004V offset on both channels (- 4mV).
In short: replacement differential pairs matched by new Sankyo; replacement Number 607 C 608 C by Nichicon mouse; Replacing the variable resistor bias adjustment for multiturn pot; and today I encouraged to take R 603 - R 604 to 23.5K.
Offset result was 45 mV in 68 mV in L and R on both -4 mV.
Bias adjustment is easy thanks to the multi-turn pot.
All thanks to your contributions. Thank you.
![IMG]](/community/proxy.php?image=http%3A%2F%2F%5Burl%5Dhttp%3A%2F%2Fi58.tinypic.com%2F14e1iyr.jpg%5B%2Furl%5D+%5B%2F+IMG%5D&hash=63554f3d4487b8493c49649528bb66df)
The Rotel RA 970BX is an excellent amplifier and should work with the best possible parameters.
I sent the post, dating back three years, but only this year I was encouraged to make the change that had advised me. Reduce the R603 / 604 resistors advised by Wahab.
http://www.diyaudio.com/forums/solid-state/213432-dc-offset-rotel.html
Finally, after three years I have decided to make a change that I had suggested.
Is to bring R 604 and R603 47k to 23k5 (placing another resistor in parallel 47k 1% metal film).
Result: -0.004V offset on both channels (- 4mV).
In short: replacement differential pairs matched by new Sankyo; replacement Number 607 C 608 C by Nichicon mouse; Replacing the variable resistor bias adjustment for multiturn pot; and today I encouraged to take R 603 - R 604 to 23.5K.
Offset result was 45 mV in 68 mV in L and R on both -4 mV.
Bias adjustment is easy thanks to the multi-turn pot.
All thanks to your contributions. Thank you.
Begins replacing the differential pair should be matched. Are the transistors Q601, Q603, Q605, Q607 in a channel, and Q602, Q604, Q606, Q608 in the other channel.
Measures the offset voltage, if not corrected continues to the next step, up to reduce R603 / 604 if necessary.
Luck.
Measures the offset voltage, if not corrected continues to the next step, up to reduce R603 / 604 if necessary.
Luck.